Table 7.
The performance and complexity comparison between the proposed CACU model vs state-of-the-art approaches.
| Crop | Model/approach name | A | MIoU | P | R | NPM |
|---|---|---|---|---|---|---|
| % | % | % | % | Million | ||
| Carrot | DeepLabV3++28 | 84.3 | – | – | – | 11.85 |
| Adapted-IV39 | 93.9 | – | – | – | 25 | |
| AgNet9 | 88.9 | – | – | 0.25 | ||
| RRUDC10 | 95.40 | – | 95.43 | 98.82 | 0.655 | |
| Proposed CAUC | 98.38 | 80.5 | 97.51 | 97.27 | 0.377 | |
| Sugar Beet | Deep encoder-decoder CNN(Bonn)53 | 94.74 | 80.1 | – | – | – |
| UNet-ResNet50(dice + focal)29 | 96.06 | 85.25 | 92.28 | 92.21 | 20.67 | |
| Bonnet26 | 99.32 | 77.47 | – | – | 1.1 | |
| UNet-ResNet5021 | – | 67.0 | – | – | 20.67 | |
| Proposed CAUC | 98.51 | 80.75 | 97.54 | 97.50 | 0.377 | |
| Sun flower | VGG-UNet31 | 90.0 | 64.0 | – | – | – |
| Bonnet21 | – | 70.0 | 1.1 | |||
| UNet-ResNet5021 | – | 43.0 | 20.67 | |||
| Bonnet26 | 99.02 | 68.98 | – | – | 1.1 | |
| Proposed CAUC | 99.10 | 81.1 | 99.14 | 99.10 | 0.377 |
A - Accuracy, P – Precision, R - Recall, S – Model Size, NPM - Number of Parameters generated by the Model, - Not given in the paper.